Navigation – Plan du site

AccueilNuméros3Measuring ethno-racial discrimina...

Sommaire - Document précédent -

Measuring ethno-racial discrimination in France: the contribution of correspondence tests

Yannick L’Horty et Pascale Petit
Traduction de Yvonne van der Does (Office of International Scientific Visibility - IdEx Université Côte d’Azur)
Cet article est une traduction de :
Mesurer des discriminations ethno-raciales en France : l’apport des testing  [fr]

Résumé

Of all the violence that plagues society, discrimination is perhaps the least visible and the most difficult to objectify. Very rarely recognized by the people who consciously or unconsciously practice it, it is strongly felt by its victims who are often unable to provide proof of a legal nature. Collecting the voices of victims through surveys or listening platforms sheds useful light on the issue, because beyond the violence and injustice felt, the feeling of discrimination influences individual choices as regards education or career and therefore social inclusion. These actions are not sufficient, however, and must be complemented by an objective and direct measure of discrimination on the perpetrators. This is the purpose of correspondence test, which has emerged in the national and international literature as the main experimental method for assessing discrimination. The aim of this paper is to present this method that has gained scientific significance and review its contribution and limitations. We will illustrate the method by drawing from an extensive overview of the twenty studies that have measured discrimination in access to employment on grounds of origin, in France, since the early 2000s. These academic studies highlight the intensity, pervasiveness, and permanence of differences in treatment based on origin in France, which make ethno-racial discrimination a major challenge for public decision-makers.

Haut de page

Entrées d’index

Keywords:

discrimination, correspondence tests
Haut de page

Texte intégral

Introduction

  • 1 “Discrimination is any distinction made between natural persons based on their origin, sex, family (...)

1Discrimination is both a criminal offense and a research topic. From a legal point of view, discrimination is a difference in treatment on the grounds of prohibited criteria, which causes prejudice in a given context such as access to employment or housing. Article 225-1 of the French Criminal Code lists more than 20 such illegal criteria.1 When proven as an offence, discrimination carries a maximum penalty of three years’ imprisonment and a fine of €45,000 if committed by a natural person and €225,000 if committed by a legal person. But from the perspective of social science research, this criminal nature is not the predominant reason for studying discrimination. The purpose of research is not either to denounce the fact that it contradicts the principle of equality and undermines both the foundations of republican values and the dignity of individuals. All this is true, but these reasons are not the basis for researchers’ interest in discrimination, which is above all a matter of gaining knowledge.

2For many sociologists, the experience of discrimination affects individuals but also has collective implications. It sheds light on what we have in common, what divides us, how we live together and how we form a society (Dubet 2014). To address the issue of discrimination is another way of studying contemporary societies and their transformations and of focusing on the processes of alienation of certain social groups. The introduction of the conceptual category of discrimination in French sociology has given a new impetus to the study of power relations, particularly those related to gender and ethnicity until now rarely considered when investigating inequality. In a French intellectual context dominated by the study of social class inequalities, sociological work on discrimination has thus enriched perspectives. Furthermore, by focusing on discriminatory processes, research on discrimination also contributes to a better understanding of how inequalities are produced (Bereni and Chappe 2011). Research is no longer limited to the sole study of discourse and ideologies, which had long prevailed when analyzing racism. Instead, it closely examines practices and interactions, even the most routine ones, which when repeated and cumulated have a powerful effect in causing inequality.

3The issue of knowledge is equally salient for economists. Their interest in discrimination stems from the fact that it appears a priori to be an irrational behavior and a market anomaly. Providers of goods and services who discriminate deprive themselves of a source of profit and are less efficient than those who treat customers or job applicants equally. According to Becker’s (1957) seminal model, discrimination should not exist in a perfectly competitive environment. Measuring and understanding discrimination is thus a way of setting limits to the classic representation in which homo economicus trades in markets of pure and perfect competition. This is undoubtedly the reason why today the economics of discrimination are a chapter in their own right of the new field of behavioral economics, which challenges the founding assumptions of the dominant economic theory (Gabuthy, Jacquemet and L’Haridon 2021).

The imperfect visibility of discrimination

  • 2 According to Heckman (1998), discrimination in the labor market occurs when an employer treats in a (...)

4A second reason that motivates researchers’ interest in the topic of discrimination is undoubtedly the difficulty of objectifying the phenomenon. Among all forms of violence, discrimination is undoubtedly one of the least visible and one of the most difficult to substantiate. From the victim’s point of view, discrimination is considered an unfair treatment and often takes the form of being refused access to a given resource (a job, housing, a public service, etc.), and the causes for this refusal are potentially multiple and difficult to identify. This type of refusal can have lasting effects on the course of a person’s life and is a violation of their dignity as a person. However, it may not be felt as such by the victim who might not be able to observe the entire process for gaining access to the resource. A job applicant, for example, does not have access to the files of other applicants. The applicant may feel discriminated against, without it being the case or not feel discriminated against when a discrimination actually occurred. From the perpetrator’s point of view, discrimination may be motivated by stereotypes that operate without the perpetrator’s knowledge. If it is unconscious, it is not in the perpetrator’s interest to reveal the fact since it is a heavily condemned offense. Moreover, the direct measure of discrimination based on pre-existing data is an impossible task. Management data or survey data, whether drawn from public statistics or not, document unfair situations, i.e. inequalities, and not differences in treatment on the grounds of legally prohibited criteria when seeking to gain access to specific resources. In application of the law of January 6, 1978, Informatique et Libertés, and the General Data Protection Regulation (GRDP), most of the prohibited criteria cannot be observed in current databases, such as company surveys or administrative data (i.e., skin color, real or supposed membership of a race or ethnic group, religious feeling, sexual orientation, disability, etc.). Even when the criteria are observable (as in the case of sex and age), the data rarely shed light on the process for accessing a resource and when they do, the data do not allow the identification of pairs of candidates applying for a resource who would have similar abilities in all respects but who would differ only on the basis of a single prohibited criterion, according to the definition of discrimination given by Heckman (1998).2 When the applicants’ data can be used (as in the case of data from civil service competitive examinations), it is nevertheless very difficult to carry out analyses in which all variables are kept constant. Researchers use sophisticated statistical technologies such as decompositions inspired by the methods developed by Blinder and Oaxaca, but these do not really allow them to observe discrimination (Bréda and Hillion 2016, Greenan et al. 2019, Challe et al. 2022).

Three mechanisms that produce discrimination

5Beyond the need to measure and produce knowledge, an additional purpose of research on discrimination is to help design and implement public actions that effectively fight against the phenomenon. Before proposing suitable actions, the first step is to interpret the causes of discrimination and identify the mechanisms at work. An additional difficulty that arises and complicates the challenge of measuring discrimination is when several mechanisms producing discrimination have been identified and each requires a specific public action.

  • 3 For studies on conceptual and empirical issues raised by research using IAT, see Corneille and Hütt (...)

6Three main theoretical mechanisms are at work in the production of discrimination. Discrimination may reflect the unconscious, unintentional stereotypes of employers who implicitly associate the applicants’ abilities with one or other of their personal characteristics (Bertrand et al. 2005). These stereotypes can be revealed using the implicit association test (IAT) proposed in social psychology (Greenwald et al. 1998).3 It may also reflect a deliberate, conscious and intentional desire on the part of the employer who either has a subjective aversion to a particular personal characteristic (this is known as preference discrimination, theorized by Becker, 1957 and corresponds to racism or sexism), or has developed a reasoning involving beliefs about the abilities of a given social group (this is discrimination based on imperfect information or statistical discrimination, following the models of Arrow, 1973 and Phelps, 1972).

7A necessary step before implementing the most suited public action is to identify these mechanisms. Awareness-raising actions may not be effective when dealing with unconscious discrimination. Stereotype awareness training is not necessarily appropriate when addressing discrimination by preference or by information. Reminding people of the legal framework and the risks of sanctions is of little use when dealing with unconscious discrimination. It should be noted that these mechanisms may also condition the measuring method: implicit association tests are of little interest in the presence of discrimination due to imperfect information.

Three specific measuring methods

8For all these reasons, a specific measuring system is required to study discrimination and the mechanisms on which it is based. Three types of sources can produce the representative data needed for these studies. The first two are based on the experience of those discriminated against and the third on the practices of those who discriminate.

  • First, data can be drawn from the spontaneous complaints reported by victims of discrimination to organizations responsible for providing support. In France, complaints lodged with the Défenseur des Droits are recorded in its annual reports. The 2020 edition of this report shows that, as part of its duties to fight discrimination and promote equality, it processed 5,196 complaints that year of a multiple nature, which represents a 15% increase since 2014. Similarly, data from the national www.antidiscriminations.fr reporting platform and its hotline (3928) indicate that in 2021, after the platform was launched on February 12, 14,000 calls were answered and 7,433 cases handled. The platform has an interface where victims and witnesses of discrimination can report cases and find support.

  • Information collected through surveys can be a second source of data. The Trajectories and Origins (T&O) survey conducted in 2008-2009 by INED and INSEE and renewed in 2019-2020 provides information on how origins are likely to affect the living conditions and career of individuals. It documents the feeling of discrimination experienced by immigrants and their descendants (Safi and Simon 2013). The Violence and gender relations survey (Virage) launched in 2015 by INED records multiple forms of violence from a gender perspective (Hamel et al. 2016).

  • The third source of data is experimental in nature and consists in directly observing the behavior of discriminators by confronting them with a pair of candidates similar in all respects except for one illegal criterion. This is the audit pair method which for a long time was based on real people and which nowadays mainly uses fictitious candidates and is known as correspondence test. Correspondence test has gradually become the reference method in the international scientific literature on discrimination, both in the labor market, which is the area most covered by studies, but also in the housing and other markets (Riach and Rich 2002, Zschirnt and Ruedin 2016, Bertrand and Duflo 2017, Quillian et al. 2017, Neumark 2018).

9Each of these data sources has advantages and disadvantages. One advantage of large victimization surveys is that they are based on samples that are representative of the general population, while over-representing minorities. However, they may be a source of non-observational errors, since persons discriminated against are not always in the best position to witness discrimination, and of non-reporting errors, since people may not wish to report themselves as discriminated against. When people do report a discrimination, their feeling of discrimination is not always an objective measure of discrimination, since they are rarely able to observe the files of other candidates and are not capable of keeping all variables constant.

Contributions and limits of correspondence test

10Field experiments also have advantages and limitations. Correspondence test, which is the most widespread form of field experiments in the literature (see du Parquet and Petit (2019) for a complete presentation of this method), consists in creating two profiles of fictitious individuals who differ only in one potentially discriminating characteristic whose effect is under examination (for example, two individuals of the same sex, age, place of residence, professional experience and education, but whose first and last name are French in the one case and North-African in the other). These fictitious individuals contact an employer in response to a formal job advertisement or spontaneously. They send the employer similar requests for information or applications. The experience is repeated with a large number of employers without their knowledge. The researchers then compare the employers’ responses to the fictitious individuals. Any significant difference in the outcome is interpreted as discrimination insofar as the only element differentiating the fictitious individuals from the employers’ standpoint is a discriminating characteristic. A variant to correspondence test is a paired audit study. It consists in hiring actors and having them meet the employers following a scenario prepared by the researchers. This approach is used to uncover situations of discrimination in face-to-face interactions. The specific limitation of this method is the possibility of appearance and personality biases that cannot be controlled by the researchers and that may distort the measure.

11Correspondence test has several advantages as regards measuring discrimination. First, researchers can perfectly control and equalize all the information about the fictitious individuals provided to the employers. Chances of success of the fictitious individuals are thus compared with all other variables kept constant, which never happens outside an experimental framework: in reality, individuals have multiple heterogeneous characteristics, which cannot be perfectly observable in the available data. Second, in real life, individuals do not all make the same efforts. This causes a selection bias that may distort the measure of discrimination. An individual may receive fewer favorable responses, for example, because the individual has contacted fewer employers. In a correspondence test, however, fictitious individuals contact the same employers the same number of times. Finally, the experimental protocol of a correspondence test allows for very precise targeting. By selecting the employers who will be tested and choosing the characteristics of the fictitious individuals, the researcher can study a large number of discrimination criteria (even those usually not identified in survey data and administrative data). Researchers can then analyze cross-discrimination and study the impact of anti-discrimination policies on eligible candidates.

12As with any assessment method, correspondence test also has its limitations, which recent studies have tended to overcome. In its standard version, correspondence test provides a non-representative measure of discrimination because it is based on a non-representative sample of data. For example, in the labor market, only a small number of occupations are usually tested because fictitious CVs are expensive. The measure of discrimination is thus partial: the identification of discrimination in one profession does not necessarily imply its presence in another profession. In addition, the data are collected over a period often limited to a few months, characterized by a particular economic situation. The measure of discrimination in hiring obtained is therefore specific to a given time. Finally, the employers tested are often concentrated in the same large geographical area, for example the wider Paris region, which means that the measure of discrimination obtained is not location-specific and cannot be associated with any local characteristics. Similar examples could be found in the housing market and in other markets where discrimination has been measured. Rather than providing representative numbers, correspondence tests serve to identify discrimination and possibly, if the protocol allows, to study its mechanisms from the perspective of the employers.

13Finally, the question of ethics arises in this type of field experiments. Tests to uncover discrimination in hiring only focus on the employers, since the candidates are fictitious. When employers are asked to choose between fictitious candidates, their behavior is examined without their knowledge. In that case, the test fails to respect the principle of fairness, according to which “the persons concerned should be informed of the treatment they will undergo. They should then be informed of the purpose of the experiment, and the experiment should be carried out accordingly (…)” (COMEDD 2010). But since discrimination is illegal, no employer would have an interest in disclosing this type of behavior knowingly. The same is true in markets other than the labor market.

14Finally, it should be noted that the protection of data collected in the context of correspondence test campaigns is now an important issue for public funders such as the French National Research Agency, which requires the drafting of demanding data management plans, anonymized data, data storage on secure servers, etc. Researchers who use this method, like all data collectors, must strictly comply with the General Data Protection Regulation (GDPR).

Convergent findings: the persistence of a high level of discrimination in France

15These limitations have long been observed and are well known. To overcome them, the solution is not to refrain from measuring discrimination and conducting tests. On the contrary, it is essential to increase the number of tests to confirm the reliability of the findings of experiments that may be partial or limited to a specific time. New knowledge can be produced by repeating the tests using different protocols. With this in mind, we have reviewed all the studies based on correspondence test in the field of access to employment using the criterion of ethno-racial origin and more precisely of North-African origin. This is the field and criterion best covered by past studies. We found 20 studies of scientific significance, based on a large number of tests (several hundred each time) that use the theory of statistical inference and regression models to confirm the validity of their results. These studies are very different as regards the professions tested, the geographical areas covered, the period in which the tests were conducted (between 2006 and 2020), and many features in their protocols distinguish them from each other. Differences could be found, more particularly, in the volume of applications sent, which ranges from 462 to 17,643. A total of 67,066 applications were sent to produce these 20 studies, an average of 3,553 per study.

16Despite these differences, we show in Figure 1 that several common findings emerge from all these studies. The figure shows the relative disadvantage suffered by North-African candidates in accessing a job interview (which corresponds to the difference in success rates compared to the success rate of the reference candidate). The 20 studies are detailed in the table in Appendix 1. The first finding is that the relative disadvantage suffered by candidates whose last name suggests a North-African origin varies greatly depending on the study. The success rate of candidates whose last name suggests a North-African origin can be from 10% lower to nearly 95% lower depending on the study. This difference can be explained by the choice of profession, employment opportunities in the geographical area, typical profiles of candidates or even different data collection periods. The second finding is that all these studies without exception confirm discrimination against North-African candidates with an average disadvantage of 41%. This discrimination affects both men and women. The third finding is that after correcting for outliers, i.e., removing the two most extreme studies, the trend does not appear to evolve over time: as a general rule discrimination is as widespread in 2020 as in 2006. All these studies show the intensity, the generality, and the permanence of ethno-racial discrimination in France.

Figure 1. Relative disadvantage experienced by applicants of North-African origin in all the published scientific studies on access to employment using correspondence test

Figure 1. Relative disadvantage experienced by applicants of North-African origin in all the published scientific studies on access to employment using correspondence test
Agrandir Original (png, 72k)

Note: The year corresponds to the year of data collection. References are provided in the bibliography.

17These key findings are confirmed by extensive reviews of the international literature. Thus, Quillian et al. (2017) point out at the end of their overview that no trend either upward or downward could be observed in discrimination. This type of conclusion may seem fragile since it is based on the compilation of the results of research in different professions, using different methodologies and in different contexts. But it is also confirmed by the only study that was conducted in France over time using an invariant methodology and applied to a specific set of professions. The study by Challe et al. (2022) as part of the Desperado research program, which has been testing the two professional categories of administrative managers and nursing assistants every two years since 2014, confirms that discrimination is persistent at a high level in France and that no declining trend can be observed.

18Another implicit finding that emerges when analyzing this figure is the increasing use of correspondence test for measuring discrimination. In the limited field of access to employment and using the criterion of ethno-racial origin, six correspondence test campaigns were conducted in the 2000s, and more than twice that number in the following decade. Throughout the 2010s, the number of tests on the housing market and tests exploring other markets has also increased. Discrimination criteria now extend far beyond ethno-racial grounds to include gender, place of residence, religion, disability, and other grounds. This considerable increase in the volume of research has been associated with numerous innovations in the correspondence test methods. Study protocols are now based on both responses to job advertisements and spontaneous applications. Researchers are capable of targeting small geographical areas, which makes spatial comparisons possible. Tests within a single company can be multiplied to produce a single company measure and pave the way to naming and shaming actions in which discriminating companies are identified and their discriminatory behavior is made public thus opening the possibility for a boycott by the population. Invariant measure protocols are repeated over time to monitor trends in discrimination. Finally, repeated correspondence test is used to evaluate public anti-discrimination policies. For example, Challe et al. (2021) examined the French free employment scheme (in which companies are subsidized for hiring employees from certain areas) and its impact on discrimination linked to the place of residence. In this case, researchers conducted a correspondence test before and after the scheme was implemented, based on the same experimental protocol. All these innovations illustrate the flexibility of correspondence test methods that pave the way for gaining new knowledge in the field of discrimination.

Haut de page

Bibliographie

Adida, Claire, David Laitin and Marie-Anne Valfort. 2010. “Identifying Barriers to Muslim Integration in France.” Proceedings of National Academy of Sciences 107(52): 384-390.

Arrow, Kenneth. 1973. “The Theory of Discrimination.” in Orley Ashenfelter and Albert Rees (eds.) Discrimination in Labor Markets. Princeton: Princeton University Press.

Becker, Gary S. 1957. The Economics of Discrimination. The University of Chicago Press.

Bereni, Laure and Vincent-Arnaud Chappe. 2011. “La discrimination, de la qualification juridique à l’outil sociologique.” Politix 94: 7-34.

Berson, Clémence. 2012. “Does Competition Induce Hiring Equity?” mimeo.

Berson, Clémence, Morgane Laouenan and Emmanuel Valat. 2020. “Outsourcing recruitment as a solution to prevent discrimination: A correspondence study.” Labour Economics 64, 101838.

Bertrand, Marianne and Esther Duflo. 2017. “Chapter 8 - Field Experiments on Discrimination.” in Abhijit Vinayak Banerjee and Esther Duflo (eds.) Handbook of Economic Field Experiments (vol. 1, pp. 309-393). Amsterdam: North-Holland.

Bertrand, Marianne, Dolly Chugh and Sendhil Mullainathan. 2005. “Implicit discrimination.” American Economic Review 95(2): 94-98.

Challe, Laetitia, Yannick L’Horty and Pascale Petit. 2022. “Le ‘testing’ montre que des discriminations persistent dans l’accès à l’entretien d’embauche y compris dans la fonction publique.” Rapport annuel sur l’état de la fonction publique, Faits et Chiffres – Dossier.

Challe, Laetitia, Yannick L’Horty, Pascale Petit and François-Charles Wolff. 2021. “Confinement et discrimination à l’embauche: enseignements expérimentaux.” Rapport d’études de la DARES 21.

Challe, Laetitia, Sylvain Chareyron, Yannick L’Horty and Pascale Petit. 2021. “Can subsidies paid directly to employers reduce residential discrimination in employment? An assessment based on serial field experiments.” Urban Studies 59(6): 1202-1218.

Challe, Laetitia, Fabrice Gilles, Yannick L’Horty and Ferhat Mihoubi. 2022, “Gender and age diversity: Does it matter for firms’ productivity?” TEPP working paper 6, http://www.tepp.eu/doc/users/268/bib/genderagediversity.pdf.

Challe, Laetitia, Yannick L’Horty, Pascale Petit and François-Charles Wolff. 2018. “Les discriminations dans l’accès à l’emploi privé et public: les effets de l’origine, de l’adresse, du sexe et de l’orientation sexuelle.” Rapport de recherche TEPP 5, http://www.tepp.eu/doc/users/268/bib/desperadoipg9376.pdf.

COMEDD. 2010. Inégalités et discriminations - Pour un usage critique et responsable de l’outil statistique, rapport au Premier Ministre du comité pour la mesure de la diversité et l’évaluation des discriminations présidé par François Héran, La Documentation Française.

Corneille, Olivier and Mandy Hütter. 2020. “Implicit? What do you mean? A comprehensive review of the delusive implicitness construct in attitude research.” Personality and Social Psychology Review 24(3): 212-232.

Dares-BIT. 2008. “Discriminations à l’embauche fondées sur l’origine à l’encontre de jeunes Français(es) peu qualifiés: une enquête nationale par tests de discrimination ou testing.” Premières informations, premières synthèses 06.3.

Du Parquet, Loïc and Pascale Petit. 2019. “Discrimination à l’embauche: retour sur deux décennies de testings en France.” Revue Française d’Économie 34(1): 91-132.

Dubet, François. 2014. “Discriminations: égalité, reconnaissance et retours du refoulé.” La nouvelle revue des sciences sociales 3: 315-349.

Duguet, Emmanuel, Noam Léandri, Yannick L’Horty and Pascale Petit. 2010. “Are Young French Jobseekers of Ethnic Immigrant Origin Discriminated Against? A Controlled Experiment in the Paris Area.” Annales d’économie et de statistique 99-100: 187-215.

Edo, Anthony and Nicolas Jacquemet. 2013. “Discrimination à l’embauche selon l’origine et le genre: défiance indifférenciée ou ciblée sur certains groupes?” Économie et statistique 464-466: 155-172.

Frémigacci, Florent, Loïc du Parquet, Pascale Petit and Guillaume Pierné. 2015. “Le mérite: un rempart contre les discriminations?” Revue française d’économie 30(1): 155-182.

Gabuthy, Yannick, Nicolas Jacquemet and Olivier L’Haridon. 2021. Économie comportementale des politiques publiques. La Découverte, coll. Repères.

Greenan, Nathalie, Joseph Lanfranchi, Yannick L’Horty, Mathieu Narcy and Guillaume Pierné. 2019. “Do Competitive Examinations Promote Diversity in Civil Service?” Public Administration Review 79(3): 370-382.

Greenwald, Anthony G., Debbie E. McGhee and Jordan L. Schwartz. 1998. “Measuring individual differences in implicit cognition: the implicit association test.” Journal of personality and social psychology 74(6): 1464-1480.

Hamel, Christelle, Alice Debauche, Elizabeth Brown, Amandine Lebugle, Tania Lejbowicz, Magali Mazuy, Amélie Charruault, Sylvie Cromer and Justine Dupuis. 2016. “Viols et agressions sexuelles en France: premiers résultats de l’enquête Virage.” Population et Sociétés 10(538): 1-4.

Heckman, James. 1998. “Detecting discrimination.” Journal of economic perspectives 12(2): 101-116.

L’Horty, Yannick. 2016. Les discriminations dans l’accès à l’emploi public, Rapport pour le Premier Ministre, La documentation française.

L’Horty, Yannick, Emmanuel Duguet, Loïc du Parquet, Pascale Petit and Florent Sari. 2011. “Les effets du lieu de résidence sur l’accès à l’emploi: une expérience contrôlée sur des jeunes qualifiés en Ile-de-France.” Économie et statistique 447: 71-95.

L’Horty, Yannick, Naomie Mahmoudi, Pascale Petit and François-Charles Wolff. 2020. “Discriminations dans le recrutement des personnes en situation de handicap: un test multi-critère”, TEPP rapport de recherche 20-5.

Manant, Matthieu, Serge Pajak and Nicolas Soulié. 2019. “Can social media lead to labor market discrimination? Evidence from a field experiment.” Journal of Economics & Management Strategy 28(2): 225-246.

Neumark, David. 2018. “Experimental research on labor market discrimination.” Journal of Economic Literature 56(3): 799-866.

Oswald, Frederick L., Gregory Mitchell, Hart Blanton, James Jaccard and Philip E. Tetlock. 2013. “Predicting ethnic and racial discrimination: a meta-analysis of IAT criterion studies.” Journal of personality and social psychology 105(2): 171-192.

Petit, Pascale, Emmanuel Duguet, Yannick L’Horty, Loïc du Parquet and Florent Sari. 2013. “Discriminations à l’embauche: les effets du genre et de l’origine se cumulent-ils systématiquement?”, Économie et statistique 464-466: 141-153.

Petit, Pascale, Mathieu Bunel and Yannick L’Horty. 2020. “Les discriminations à l’embauche dans la sphère publique : effets respectifs de l’adresse et de l’origine.” Revue économique 71(1): 31-56.

Phelps, Edmund S. 1972. “The statistical theory of racism and sexism.” The american economic review 62(4): 659-661.

Pierné, Guillaume. 2013. “Hiring Discrimination Based on National Origin and Religious Closeness: Results from a Field Experiment in the Paris Area.” IZA Journal of Labor Economics 2(4): 1-15.

Quillian, Lincoln, Devah Pager, Ole Hexel and Arnfinn H. Midtbøen. 2017. “Meta-analysis of field experiments shows no change in racial discrimination in hiring over time.” Proceedings of the National Academy of Sciences 114(41): 10870-10875.

Riach, Peter A. and Judith Rich. 2002. “Field Experiments of Discrimination in the Market Place.” The Economic Journal 112(483): F480-F518.

Safi, Mirna and Patrick Simon. 2013. “Les discriminations ethniques et raciales dans l’enquête Trajectoires et Origines: représentations, expériences subjectives et situations vécues.” Economie et statistique 464(1): 245-275.

Valfort, Marie-Anne. 2019. “La religion, facteur de discrimination à l’embauche en France?” Revue économique 68(5): 895-907.

Zschirnt, Eva and Didier Ruedin. 2016. “Ethnic discrimination in hiring decisions: A meta-analysis of correspondence tests 1990-2015.” Journal of Ethnic and Migration Studies 42: 1115-1134.

Haut de page

Annexe

Table. Correspondence tests carried out in mainland France, on the labor market and the housing market, using the criteria of ethno-racial origin

Market

Period and place of collection

Discrimination criteria examined

Field

Number of profiles

Sample size

Main significant results

Labor

“End 2005
to mid-2006”

6 employment areas:

Lille, Lyon, Marseille,

Nantes, Paris and Strasbourg

Origin (French, North-African or Sub-Saharan African)

Grouping by sector: hotel and catering, trade, construction and services

6 candidates: 3 men and 3 women of French, North-African or sub-Saharan African origin

2,440 job openings tested

4,880 applications sent (2 per job, of the same sex, one of French origin, the other of North-African or sub-Saharan African origin)

Discrimination against candidates of foreign origin, stronger among men and against candidates from sub-Saharan Africa

Labor

Spring 2009

Religion (Christian or Muslim) and origin (French or Senegalese) signaled by a Christian or Muslim first name and whether the candidate’s participation in a religious scouting organization

Secretaries and accounting assistants

3 female candidates, with or without photograph on the application

271 job openings tested

542 applications sent (two per job)

The female candidate of French origin had the same chances of obtaining an interview as the female candidate of either Muslim or Christian African origin

Discrimination against the female Muslim candidate is significantly stronger than discrimination against the African female Christian candidate. The presence of photographs does not change the results.

Labor

October to December 2006

Paris region

Nationality (French or Moroccan), origin (French or North-African) and place of residence

Accountants

8 male candidates

164 job openings tested

1,312 applications sent (8 per job)

Discrimination against foreign male candidates or those of foreign origin. Discrimination based on the place of residence of the male candidate of French origin

Labor

December 2008 to January 2009

Paris region

Sex, origin (French or North-African) and place of residence (Villiers-le-Bel, Sarcelles, Enghien-les-Bains)

IT developers

12 male candidates

307 job openings tested

3,684 applications sent (12 per job)

Discrimination based on the place of residence against female candidates of French origin

Disadvantage specific to residence in Villiers-le-Bel rather than Sarcelles against the female candidate of French origin

Labor

July to November 2011

Origin (French or North-African) and sex

Supermarket cashier

2 male candidates

1,250 stores tested in 54 employment areas

Discrimination based on origin that does not depend on the store’s exposure to competition

Discrimination in favor of women, which is reversed in cases of low exposure to competition

Labor

September 2011 to February 2012

Paris region

Sex, origin (French, North-African or non-identifiable foreign), based on a signal of French language proficiency

Accounting sector

6 candidates distinguished by sex and origin

504 job openings tested

3,064 applications sent (6 per job)

Discrimination based on origin regardless of sex and origin

The language proficiency signal eliminates discrimination based on origin for women

Labor

February to April 2009

Paris region

Sex and origin (French, North-African, Vietnamese or Senegalese)

IT developers

8 candidates: 4 women and 4 men

303 job openings tested

2,422 applications sent (8 per job)

Discrimination in favor of women among candidates of Vietnamese origin

Women of North-African or Senegalese origin suffer a double disadvantage related to their sex and origin; they are less often contacted for an interview, and when they are, they are more often contacted after the other candidates

Labor

April to September 2011

Paris region

French or North-African origin and participation in a secular association or an association with a Catholic or Muslim signal

Real estate agents

6 male candidates

300 job openings tested

1,800 applications sent (6 per job)

Discrimination based on origin, irrespective of the religious signal

Discrimination based on the Muslim religion irrespective of the candidate’s origin

Labor

April to September 2011

Paris region

Origin (French or North-African), mention of an award in the French competition for best apprentice

Masons, plumbers, electricians

4 male candidates

301 job openings tested

1,204 applications sent (4 per job)

Discrimination based on origin, regardless of the applicant’s abilities

Mention of a prize in the competition offsets the disadvantage based on the North-African origin of one candidate, suggesting a Becker-type discrimination

Labor

October to December 2006

Paris region

Nationality (French or Moroccan), origin (French or North-African) and place of residence

Waiters

8 male candidates

160 job openings tested

1,280 applications sent (8 per job)

Discrimination against foreign candidates or those of foreign origin

Discrimination based on the place of residence of the candidate of French origin

Labor

April to July

All of France

Origin

Bank and insurance manager, sales rep, bank and insurance technician, etc.

2 fictitious candidates of different origin (French and North-African)

1,433 tests conducted

2,866 applications sent

Significant discrimination against North-African applicants

Labor

September to December 2014

Mainland France

Catholic, Jewish or Muslim religion signaled by the first name, education in a faith-based middle school and participation in a faith-based scouting organization

Accountants, assistant accountants, accounting administrative assistants

30 candidates: 24 French candidates born in Lebanon distinguished by their sex, their supposed religion, the intensity of the signal of religious practice and the quality of their application and 6 candidates for "confirmation"

6,231 job openings tested

6,231 applications sent (one per offer)

Discrimination against Jews and Muslims. Muslim men face the most discrimination, and discrimination increases, when a signal of religious practice is added

The greater quality of the application eliminates discrimination against female Jewish and Muslim candidates but increases discrimination against male Muslim candidates

Labor

October 2015 to April 2016

Paris region

Origin (French or North-African), place of residence (priority district or not)

Administrative managers, maintenance technicians, nurse assistants in the public and private sectors

3 candidates of undetermined sex in the profession tested

1,086 job openings tested

3,258 applications sent (3 per job)

Discrimination is not lower in the public sector

Discrimination in the hospital public service and the territorial public service

Discrimination in the recruitment of public servants in the territorial public service

Labor

March 2012 to March 2013

Paris

Origin (French and North-African, only visible on Facebook).

Accountants

2 fictitious candidates with a French-sounding first and last name, or born either in Brive la Gaillarde or Marrakech as shown on their Facebook page exclusively

462 job openings tested

230 by the candidate of French origin and 232 by the candidate of North-African origin

Significant discrimination based on origin, based in part (37%) on the fact that employers check Facebook pages

Labor

October 2018 to January 2019

Ile de France, Lyon, Toulouse, Nice, Bordeaux and Nantes

Origin (French or North-African) and place of residence (priority district or not), by gender.

Applications to job openings: sales assistant, account manager, IT developer, accountant, administrative manager, waiter.

Request for information about job openings: all professions

Spontaneous applications: receptionist and maintenance technician.

3 fictitious candidates of the same sex distinguished by their origin and place of residence.

40 SBF 120 companies.

Total: 17,643 observations

Multi-channel approach:

Applications in response to job openings, requests for information in response to job openings, non-formal spontaneous applications (without CV), non-formal spontaneous applications (with CV).

Discrimination based on origin in all test areas

Less significant discrimination based on the place of residence

Absence of significant discrimination based on the place of residence at company level, but within the industry and Paris region

Labor

5 series of correspondence tests: between the end of 2015 and the end of 2020

Origin (French or North-African) and place of residence (priority district or not)

Administrative managers (private and public sector)

3 fictitious male individuals.

1,583 jobs tested

4,749 applications sent

While occasional and partial discrimination was observed before the lockdown in spring 2020, none was detected during the lockdown

Conversely, a sharp increase in discrimination related to origin and place of residence was observed after the lockdown in the fall of 2020

Overall, the fluctuations in discrimination reflect those in unemployment

Labor

Assessment of the French free employment scheme

Experimentation with the free employment scheme from April 2018 in nearly 200 priority districts.

3 correspondence test campaigns every 6 months between February 2018 and May 2019.

Origin (French or North-African) and place of residence (priority district or not)

Waiters, account managers, accountants.

4 fictitious male candidates distinguished by their origin (French or North-African) and their place of residence (priority district or not).

2,436 job openings tested

9,744 applications sent

Significant discrimination based on origin

Less significant discrimination based on the place of residence

Deployment of the French free employment scheme did not coincide with a change in discrimination linked to origin or place of residence as this result is not linked to employers’ lack of information about the scheme.

Labor

October 2019 to March 2020

Paris region

Origin (French or North-African), sex, place of residence (city priority district or not) and disability

Administrative manager, nurse assistant

5 fictitious candidates of different sex, origin (French and North-African), and with a disability

463 job openings tested

2,315 applications sent

No discrimination based on sex or address.

Confirmed discrimination based on North-African origin

Labor

October 2020

Avesnois

Origin (French or North-African) and reputation of the place of residence in Maubeuge according to their sex.

Administrative

4 fictitious candidates of different sex, origin (French or North-African) and place of residence of varying reputation in Maubeuge (priority district or not).

1,000 companies representative of the employment area

2 spontaneous applications of the same sex (a candidate of French origin residing in a non-priority district of Maubeuge versus a candidate of French origin residing in a priority district of Maubeuge or a candidate of French origin residing in a non-priority district of Maubeuge versus a candidate of North-African origin residing in a non-priority district of Maubeuge)

Absence of significant discrimination based on origin. Discrimination based on the reputation of the place of residence for men only.

Note: All the studies presented in this table are based on correspondence tests. One exception is the DARES-BIT study (2008) which collected the data by sending written job applications, but also by responding by phone or in person at the company. Moreover, in all the studies that examine the effect of origin, the sound of the candidate’s first name and last name signals the origin. One exception is the DARES-BIT study (2008) in which the North-African origin of fictitious candidates is signaled not only by the sound of the first and last name but also by the skin color of the candidate who responds in person at the company. Similarly, Adida, Laitin and Valfort (2010) compared the rate at which candidates were contacted for an interview with or without photographs in their application.

Haut de page

Notes

1 “Discrimination is any distinction made between natural persons based on their origin, sex, family status, pregnancy, physical appearance, particular vulnerability resulting from their economic situation, apparent or known to the perpetrator, name, place of residence, state of health, loss of autonomy, disability, genetic characteristics, morals, sexual orientation, gender identity, age, political opinions, trade union activities, ability to express themselves in a language other than French, membership or non-membership, real or supposed, of an ethnic group, a nation, an alleged race or a specific religion” (Art. 225-1, penal code).

2 According to Heckman (1998), discrimination in the labor market occurs when an employer treats in a different way two individuals with identical productivity but who belong to different demographic groups.

3 For studies on conceptual and empirical issues raised by research using IAT, see Corneille and Hütter (2020) and Oswald et al. (2013).

Haut de page

Table des illustrations

Titre Figure 1. Relative disadvantage experienced by applicants of North-African origin in all the published scientific studies on access to employment using correspondence test
Légende Note: The year corresponds to the year of data collection. References are provided in the bibliography.
URL http://0-journals-openedition-org.catalogue.libraries.london.ac.uk/alterites/docannexe/image/473/img-1.png
Fichier image/png, 72k
Haut de page

Pour citer cet article

Référence électronique

Yannick L’Horty et Pascale Petit, « Measuring ethno-racial discrimination in France: the contribution of correspondence tests », Appartenances & Altérités [En ligne], 3 | 2023, mis en ligne le 01 mars 2023, consulté le 26 mars 2025. URL : http://0-journals-openedition-org.catalogue.libraries.london.ac.uk/alterites/473 ; DOI : https://0-doi-org.catalogue.libraries.london.ac.uk/10.4000/alterites.473

Haut de page

Auteurs

Yannick L’Horty

Université Gustave Eiffel, ERUDITE (EA 437), TEPP (FR 2042), F-77454, Marne-La-Vallée, France, yannick.lhorty@univ-eiffel.fr

Pascale Petit

Université Gustave Eiffel, ERUDITE (EA 437), TEPP (FR 2042), F-77454, Marne-La-Vallée, France, pascale.petit@univ-eiffel.fr

Haut de page

Traducteur

Yvonne van der Does (Office of International Scientific Visibility - IdEx Université Côte d’Azur)

Haut de page

Droits d’auteur

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Sommaire - Document précédent -
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search